Geena Davis has spent her career playing complicated, powerful women onscreen--and working behind the scenes to address Hollywood's endemic diversity problems. In 2004, the Oscar-winning actor founded the Geena Davis Institute on Gender in Media, a nonprofit research organization that tracks onscreen portrayals of women and girls. Nearly 15 years later, as her industry reckons with #MeToo and Time's Up, Davis is continuing to advocate for change--and to collect the data that makes a difference.

Her latest target: the pitifully low numbers of onscreen women in science, technology, engineering, and mathematics, or STEM. A recent report published by the Geena Davis Institute, funded by Lyda Hill Philanthropies, surveyed the past 10 years of popular films and TV shows and found some grim numbers: Almost 63 percent of fictional characters shown working in STEM are men, and more than 71 percent are white.

"As abysmal as the numbers are in real life for women, it's far worse in fiction--where they make it up!" says Davis, half-amused and half-exasperated. In the following condensed and edited interview, Davis discusses why fictional role models matter; why she's generally avoided public confrontation when advocating for gender diversity; and how she picks her acting roles these days--including an upcoming turn on Netflix's Glow.

Why did you start the Institute, and how did it go from idea to reality?

When my daughter was 2, I sat down to watch kids' videos with her. I immediately noticed there were far more male characters than female characters--in what we're showing kids in the 21st century! We're training kids to have gender bias from minute one. I was appalled.

At first, I didn't intend to make it my life's mission; I just started bringing it up whenever I had a meeting with a director, producer, or studio executive. I'd say, "Have you ever noticed how few female characters there are in movies made for kids?" Every single person said, "Oh, no, that's not true anymore. That's changed." They would very often name a movie with one female character as proof that gender inequality had been fixed.

They all think it's not a problem--so I wanted the data. That's when I immediately jumped to sponsoring the largest research study ever done on gender differences in kid's TV and movies. The research now covers more than a 20-year span. It's right there in black and white. No one ever has any room to doubt that this is accurate information.

What's the biggest impact you've seen from your work at the Institute?

We were thrilled to find recently that [the percentage of] family films with a female lead character has doubled in the past four years. We have also found that for the past four years in a row, family movies starring a female character have made significantly more profit at the box office. It's not only the moral imperative now, but also the financial end.

With so many diversity problems in Hollywood, why focus specifically on onscreen representation of women in STEM?

STEM has been incredibly important to us from the beginning. When we first started looking at gender depictions in kids' media, one of the big things that we looked at was occupations of female characters. Currently in "family-rated" films--what we consider G, PG, and PG-13-rated--81 percent of the jobs are held by male characters. Whereas women are 51 percent of the population! And [nearly] 50 percent of the workforce in America, 40 percent globally. It's widely skewed.

So as abysmal as the numbers are in real life for women, it's far worse in fiction--where they make it up! It's not like they can even claim, "Well, we're hampered by life. We feel we need to reflect reality in fiction." They're not. They're creating their own worse-than-the-tragic-reality.

Right--and there's no pipeline problem in fiction.

Exactly, exactly! This is what I point out constantly: The easiest thing in the world is to push beyond reality. Make the president's cabinet half-female. Make the board half-intersectional-female [characters]. But the default [in fiction] is worse than reality.

My contention is that because we have so few real-life role models in every field, we need fictitious role models. It has the same impact. Seeing somebody do something in real life is incredibly important, but it can be equally important to see a character onscreen doing things.

Your recurring role on Grey's Anatomy was a woman in STEM, a badass surgeon. Do you pick your roles by thinking about what you want to represent?

I wish I had such a wealth of roles pouring in that I could cherry-pick ones that would be good role models! That's not the case. However, ever since Thelma & Louise, I do put a lot of thought into the parts I choose. That movie made me realize how few opportunities we women have to feel empowered by and excited about female characters. For men, literally every movie they watch they can identify with some character and live vicariously through them, but it's actually very rare for women.

I always try to keep in mind what the women in the audience are going to think about my character. But I avoid thinking in terms of "role models," because I feel like that term can be limiting in some ways. If you think about Thelma and Louise, we're the worst role models in the world! I like characters who are deeply flawed, but rise above and find their strengths.

What's your next big project, whether for the Geena Davis Institute or your day job?

I just finished shooting on a movie with Jessica Chastain called Eve, and I've joined the cast of that Netflix series Glow.

Really? I love that show. Whom are you playing?

I'm doing five episodes this season with them. I just started that, and it was super fun. I think the plot line is secret at this point. I'm not [playing] a wrestler or a former wrestler.

You've previously talked about the dearth of roles you got offered once you hit 40. Have you seen that changing at all recently?

No, those numbers haven't moved. Most numbers in this industry do not move, and that's the problem. Female directors have been stuck in the low single digits for decades. For all the talk and attention they get, the numbers never budge.

However, I will say that I feel like I landed on something that's the lowest-hanging fruit. The easiest, quickest thing to change is onscreen representation in children's TV and movies. That bias is completely unconscious. People making children's entertainment had no idea they were leaving out that many female characters until we had the data to prove it to them--and people making entertainment for children care about children and want to do right by them. There's a built-in desire to do right.

So what, if any, changes have you seen as a result of #MeToo and Time's Up?

I think we really are at the beginning phase of a new wave of the women's movement. It is so freeing for women to know that now it's OK to talk about this stuff, [instead of feeling] like, "Well, I don't want people to think I'm difficult to work with" or "I don't want to cause a problem." When Gillian Anderson found out David Duchovny was making twice as much as she was [for The X-Files revival], bam, she goes public with it. That was just plain unheard-of. So many of my peers and friends are so outspoken about all of this now. Think of all the incredible women leading the way in this space.

I've just been quietly doing this privately. I take very private, collegial, "Hey, I bet you didn't realize" kinds of meetings. But there's a whole other level now of profound outspokenness, which is just very exciting and I think will create lasting change.

That private, collegial, "Hey, I bet you didn't realize" tactic versus calling public attention to inequity--is one approach more effective than the other?

I think the best way to approach unconscious bias is to say, "You didn't know this. This is not me standing and blaming you. This is pointing out something you did not know. Can we work together to make this better?" That works great in that instance.

For conscious bias, all bets are off. You have to confront. You can't quietly say, "I know you're deliberately holding women back, but can I politely ask that you don't?" That doesn't work. So for different circumstances, we need different approaches.